1. Field of the Invention
The present invention relates to hot swapping of firmware provided in storage units connected to host apparatuses.
2. Description of the Related Art
A redundant array of inexpensive disks (RAID) system includes a server and a RAID unit connected to the server. A plurality of servers can be connected to a RAID unit. In hot swapping of firmware of a RAID unit, when a plurality of control modules (hereinafter described as CMs) that include the firmware exist, the CMs are separated into a first half group and a second half group, and hot swapping is performed separately for the first and second half groups. For example, in the case of a RAID unit that includes four CMs, firmware of two of the CMs is first replaced at the same time. Then, firmware of the other two of the CMs is replaced. A multipath configuration is adopted for the RAID unit and the server, and individual paths are connected to the corresponding CMs. Thus, the firmware can be hot-swapped without stopping the service. The cases of hot swapping of firmware in a RAID unit that includes two CMs will now be described.
FIG. 1A shows a case where known swapping of firmware is normally completed. A server and a RAID unit that includes two connection paths connected to the server are provided.
(1) In a regular operation, data is transferred from the server to the RAID unit using the two paths.
(2) During swapping of firmware of a first half part CM0 of CMs, data is transferred between the server and the RAID unit using only one of the paths on the side of a second half part CM1 of the CMs.
(3) During swapping of firmware of the second half part CM1, data is transferred between the server and the RAID unit using only one of the paths on the side of the first half part CM0, swapping of the firmware of which has been completed.
(4) When swapping of the firmware of the second half part CM1 is completed, data is transferred using the two paths.
FIG. 1B shows a case where the known swapping of firmware is abnormally terminated. In a case where something is wrong with one of the paths, when hot swapping of firmware is performed using the other normal one of the paths, the service is stopped because all the paths between the RAID unit and the server cannot be used as paths for transactions. In this case, at the time swapping of the firmware of the first half part CM0 is started, there is a problem with connectivity of the path connected to the second half part CM1.
(1) A problem occurred with the path connected to the second half part CM1 before swapping of firmware, and thus data is transferred using only the first half part CM0.
(2) When swapping of the firmware of the first half part CM0 is started, data cannot be transferred through the path on the first half part CM0 side.
(3) Since data cannot be transferred through both of the paths, the server detects an error, so that the service is stopped. The same applies to a case where, at the time swapping of the firmware of the second half part CM1 is started, there is a problem with connectivity of the path connected to the first half part CM0.
In a case where the service may be stopped when firmware is hot-swapped, as described above, it is required that swapping of the firmware be not performed at the time, and swapping of the firmware be performed after the cause of the problem with the path is eliminated.
Thus, in the known method, before the firmware of the first half part CM0 is replaced, a maintenance person logs in the server and issues commands to check connectivity of the paths between the RAID unit and the server. When there is no problem, the firmware of the first half part CM0 is replaced. Moreover, before the firmware of the second half part CM1 is replaced, a similar operation is performed.